SpaceRay
![]() |
I have found this image shown below, and is all black and white, and I wonder if there could be a way to make a filter that could be able to fill each of the enclosed shapes with different random colors, or at least with different random grayscale
i know that FF does not have any Flow fill component, and maybe I am asking something impossible, but I wonder if there could be some possible way that FF could recognize the edges and fill the inside of them Thanks very much The source image shown here is made with Processing.org software W Blut Voronoi ![]() |
|||||||||||
Posted: December 17, 2015 9:45 am | ||||||||||||
SpaceRay
![]() |
||||||||||||
Posted: December 17, 2015 9:53 am | ||||||||||||
SpaceRay
![]() |
or also as said it could be also instead filling with differents shades of gray
![]() |
|||||||||||
Posted: December 17, 2015 10:04 am | ||||||||||||
SpaceRay
![]() |
Or if you know any other software that could be able to do this with a GUI based thing without scripting or programming as it happens with processing
|
|||||||||||
Posted: December 17, 2015 10:06 am | ||||||||||||
Zoltan Erdokovy
![]() |
Take a look at this thread: https://www.filterforge.com/forum/read...&TID=13353
Apparently this is a tricky problem. EDIT: Lol, sorry, you actually know about that thread. ![]() |
|||||||||||
Posted: January 10, 2016 2:21 pm | ||||||||||||
Skybase
![]() |
SpaceRay, I don't know if you ever received my PM via FilterForge but I dropped you the possible solution via other programs there.
Either way for everybody else: the problem is that it's extremely tricky, especially when you don't have the initial data necessary to produce a filling effect. So for example, you can probably have an image with only the outlines of each cell. The issue here is that there's no data telling anybody where to get the colors from. So the question becomes, how do you make that data from the current data set? It seems like the only program so far that does this SEEMS to be the one stop solution is Genetica. |
|||||||||||
Posted: January 10, 2016 10:51 pm | ||||||||||||
SpaceRay
![]() |
thanks skybase and Zoltan, Yes I know that thread thanks
I know it may be a very difficult thing for the software to be able to find the edges of the enclosed shapes and fill them with different colors, and there is no data reference to know where are all these enclosed shapes, and they are all irregular and different, so there is no pattern to follow. The hand made filling exist since Deluxe Paint in 1985, but this depends on that yourself give the starting point already inside one of the enclosed shapes, and will only detect the edges of only that only shape i agree that is really extremely tricky and surely is needed some very complex scripts or coding to be able to recognize shapes and able to fill them automatically
no, I did not receive it, sorry, maybe I should look for it
Can this be done in Genetica? how do you it? i have Have Genetica 4.0 but regretably I still can not understand how to build things with it, only use it for seamless tiling |
|||||||||||
Posted: January 11, 2016 12:46 am | ||||||||||||
Skybase
![]() |
In the PM I pointed to region selection feature unique to Genetica. At the basic level, this is an extremely extremely unique feature that I really don't know what other program has. Which to me speaks of how complicated the process must be.
The construction techniques are much very similar to FilterForge. Data is streaming from one point to another through nodes. I suggest messing with it. The specific feature you're looking for is regions. Again, I've never seen anything like it anywhere in the industry. This is extremely "genetica" unique. |
|||||||||||
Posted: January 11, 2016 4:54 am | ||||||||||||
Indigo Ray
![]() |
Maybe the stuff we did recently in this thread with the bomber is a start? It looks similar to the example picture from Genetica, but is not the same. It seems that the "regions" node in Genetica can find regions of any input (including an image!), whereas our filters only work with regions generated by the bomber component.
|
|||||||||||
Posted: January 11, 2016 8:13 pm | ||||||||||||
ThreeDee
![]() |
I believe we've covered this before: It is possible with scripting, but due to FF being procedural, not pixel-based, it will be extremely inefficient -- hundreds, perhaps thousands of times more inefficient than in a pixel-based programming environment where one can give new values to pixels as you go. With FF you have to either use large tables to store the intermediate results and risk running out of memory or do thousands of redundant calculations for the same pixels and risk running out of patience.
|
|||||||||||
Posted: January 12, 2016 2:12 pm | ||||||||||||
Rachel Duim
![]() |
Thanks ThreeDee. Right on the money. This point (used up resources due to the procedural nature of FF) shows up elsewhere (with or without LUA): Fractals, reaction diffusion, anti-aliasing (FXAA) among many other examples. Any concept that requires a large bitmap and multiple passes though the data are going to be slow. I know, I have coded fractals and reaction diffusion with the resulting time to go to dinner
![]() Math meets art meets psychedelia. |
|||||||||||
Posted: January 12, 2016 3:30 pm | ||||||||||||
ThreeDee
![]() |
That out of the way, an interesting potential approach to the problem occurred to me.
The filter only has to do one thing: create a grayscale image with a different index value for each region. What if the procedure were to be broken down to two passes: 1) Find the regions inside each rendering thread only -- either an 8 x 8 grid or 16 x 16 grid, which is how FF break the image down for multithread rendering -- and give each region an index value. Apply and save this file. 2) Open this intermediate file and use a second pass filter to find any border areas where two index values should be the same and make them so. Both steps would have to be scripted, but this would be at least 64 or 256 times more efficient than trying to do it all in one go. Just saying. |
|||||||||||
Posted: January 12, 2016 4:36 pm | ||||||||||||
Skybase
![]() |
I kinda think the whole situation of needing to fill a lot of empty regions for whatever reason is kinda niche in its own way. Usually when you're confronted with a complicated task, you have ways of handling that from the start. In such a case you'd construct things so that you don't run into problems like that. So to me, stuff like "Now I need to fill holes with random colors (i.e.)" is an after-thought.
Example image from Genetica using the image above. Damn. That's impressive. It's a bit messy but it works. The anti-alias clearly got in the way of Genetica but I guess you can reapply anti-alias later using some other tool. It even nailed the little tiny areas. [Edit] Then again I'm wondering why go through the trouble lol. ![]() |
|||||||||||
Posted: January 12, 2016 7:35 pm | ||||||||||||
ThreeDee
![]() |
Which is the essential answer as to why no one has done this with FF. |
|||||||||||
Posted: January 13, 2016 7:46 am | ||||||||||||
SpaceRay
![]() |
As I knew that would be complex and difficult in FF, is why I wrote this as an alternative, and so it seems that the alternative is Genetica 4.0
WOW, WOW! How have you done that with Genetica??? ![]() It looks good and interesting, well done, congratulations How many colors can be used and how are they distributed?
Thanks for explaining in this good way why is very complex and difficult to do it in FF, I understand it. But then you have wrote a possible alternative way to do try to find a way to do it in some way?? ![]()
i have no idea how to make this, would this be possible and work in some way?
This happens if YOU are the one that is creating yourself that black and white shape, that you know that you will need to fill the regions with colors and have to plan ahead how you are going to do it, and not do it and then think later how you are going to do it, so I agree that you have to think about that from start But if you have a software that is able only to draw the black and white shapes as it happens with FF but is not possible to fill them because this software does not have any possible way to fill that regions and holes, you have to search for another posible solution. |
|||||||||||
Posted: January 14, 2016 1:37 am | ||||||||||||
Skybase
![]() |
Oh.... it was kinda super simple. Just hook up the region node. Took me 2 minutes to figure out. I know nothing about Genetica in general. Check the link out, it's been mentioned here on this forum by IndigoRay: http://spiralgraphics.biz/genetica/he...egions.htm
Exactly, the part where you continuously tack on solutions back to back is kinda where the whole issue stems fr om. You can see how it becomes more complicated and more costly as you add more solutions back to back. i.e. [Processing output] ---> [Genetica regions]---> [anti-alias filter process???] But I sometimes think it's the only solution, especially when you just lack the resources or skills. And I think if it gets you somewh ere it's fine. Just know that there's always a better way. Speaking of stuff I've been working on making something very similar to the picture above. This was done using FilterForge. It's different in many ways but I guess it can be adapted. PM me for filter if you wish. ![]() |
|||||||||||
Posted: January 14, 2016 1:53 am | ||||||||||||
SpaceRay
![]() |
Beautiful and lovely colorful example you have here in the post above, like it very much, and the color compositions and distribution is well done and balanced, and is well done the random that is not repeated much on near shapes, good.
Oh, it seems that it may be simple to use. I will try to see how it works, thanks for the link. |
|||||||||||
Posted: January 14, 2016 2:20 pm | ||||||||||||
SpaceRay
![]() |
Finally I have just found the PM email you sent, thanks very much for explaining it and the suggestions. |
|||||||||||
Posted: January 15, 2016 2:50 am | ||||||||||||
Skybase
![]() |
Spaceray, I got the PM and sent a response to the email listed for that private message about a week back. This forum message is the only way it seems I can reach you for now. I got another message today so I thought I'd notify here.
|
|||||||||||
Posted: January 22, 2016 5:25 am |
Filter Forge has a thriving, vibrant, knowledgeable user community. Feel free to join us and have fun!
33,719 Registered Users
+8 new in 7 days!
153,545 Posts
+5 new last day!
15,348 Topics
+71 new in year!
33 unregistered users.